ERROR:root:Could not allocate bytes object! about stable-diffusion-webui-tensorrt HOT 12 CLOSED

This sounds like torch is running out of system memory. How much RAM do you have installed in your system? Can you retry with as few apps running as possible?

from stable-diffusion-webui-tensorrt.

This sounds like torch is running out of system memory. How much RAM do you have installed in your system? Can you retry with as few apps running as possible?

i have 16gb ddr5 and tried to run SD only, still had the same issue

from stable-diffusion-webui-tensorrt.

Can you open task manager and monitor the memory use? At this point, this is my best guess..

from stable-diffusion-webui-tensorrt.

Can you open task manager and monitor the memory use? At this point, this is my best guess..

yea... it exceeded 95% for few secs and then drop down
but now it crash instead
could be the onnx tho idk

from stable-diffusion-webui-tensorrt.

yeah i have/has the same issue, 32GB DDR4 8GB VRAM,

for some reason after creating for than 1 engine with "Export to TRT" python just wanted 25GB RAM allocated, after killing the "python.exe" which alone had 10GB of the RAM, restarting the webui and it worked

from stable-diffusion-webui-tensorrt.

so wondering if there's anyway to hard reduce the ram usage with the draw back of slower engine building?

from stable-diffusion-webui-tensorrt.

You could try setting diable_optimizations = True to reduce the VRAM consumption here: https://github.com/NVIDIA/Stable-Diffusion-WebUI-TensorRT/blob/main/exporter.py#L99

from stable-diffusion-webui-tensorrt.

You could try setting diable_optimizations = True to reduce the VRAM consumption here: https://github.com/NVIDIA/Stable-Diffusion-WebUI-TensorRT/blob/main/exporter.py#L99

from stable-diffusion-webui-tensorrt.

Do not modify the if statement and just add a line above it as shown here:

diable_optimizations = True
if diable_optimizations:
 .....

from stable-diffusion-webui-tensorrt.

Do not modify the if statement and just add a line above it as shown here:

diable_optimizations = True
if diable_optimizations:
 .....

well thank you it actually work now~
is it possible to document this just in case someone gets the same issue?

from stable-diffusion-webui-tensorrt.

There is a chance this breaks things down the line when using LoRA. I'd need to test that first before recommending it.

If this doesn't break functionality, I could ad an "lowram" option in the UI.

from stable-diffusion-webui-tensorrt.

16gb vram and 16gb ram, ram was the issue

simply adding 32gb of pagefile/swap fixed the issue though

from stable-diffusion-webui-tensorrt.